Latest working kernel version: 2.6.22.14 Earliest failing kernel version: 2.6.24.2 Distribution: Ubuntu-server 7.10 Hardware Environment: HP Proliant ML370 G3 Software Environment: x86 Problem Description: CCISS module hangs on boot. Compiling the module into the kernel makes no difference. Steps to reproduce: Use 2.6.24.2 on a system with HP Smart Array 6402 Last kernel messages: Ramdisk driver initialized: 16 RAM disks of 65536K size 1024 blocksize HP CISS Driver (v 3.6.14) ACPI: PCI Interrupt 0000:02:01.0[A] -> CSI 16 (level, low) -> IRQ 16 cciss0: (0x46) at PCI 0000:02:01.0 IRQ 223 using DAC blocks= 35553120 block_size= 512 heads=255, sectors=32, cylinders=4357 blocks= 213367680 block_size= 512 heads=255, sectors=32, cylinders=26140 NOTE: At first boot I was dropped to initramfs when the kernel failed to find /. Part of dmesg from working 2.6.22-14-server kernel (ubuntu-server): [ 103.126800] cciss0: <0x46> at PCI 0000:02:01.0 IRQ 19 using DAC [ 103.156797] blocks= 35553120 block_size= 512 [ 103.166762] heads=255, sectors=32, cylinders=4357 [ 103.166765] [ 103.176762] blocks= 213367680 block_size= 512 [ 103.186760] heads=255, sectors=32, cylinders=26148 [ 103.186763] [ 103.187358] blocks= 35553120 block_size= 512 [ 103.187479] heads=255, sectors=32, cylinders=4357 [ 103.187481] [ 103.187485] cciss/c0d0: p1 p2 [ 103.191958] blocks= 213367680 block_size= 512 [ 103.192073] heads=255, sectors=32, cylinders=26148 [ 103.192074] [ 103.192077] cciss/c0d1: p1 [ 103.377089] ata1.00: ATAPI: COMPAQ CD-ROM SC-148C, B101, max PIO4 [ 103.421982] Attempting manual resume Other notes: The problem might be similar to bug #9859, I too compiled with make-kpkg, could this be causing the problem? I tried setting up netconsole for full output, but no luck.
Ok, it seems I get droppet to initramfs everytime after some minutes. Here's some more info from the failing kernel: [ 151.916425] blocks= 35553120 block_size= 512 [ 151.926421] heads=255, sectors=32, cylinders=4357 [ 151.926423] [ 151.936423] blocks= 213367680 block_size= 512 [ 151.946416] heads=255, sectors=32, cylinders=26148 [ 151.946417] [ 152.075507] ata1.00: ATAPI: COMPAQ CD-ROM SC-140C, B101, max P104 [ 152.256729] ata1.00: configured for P104 [ 152.416390] scsi: waiting for bus probes to complete ... [ 155.577190] scsi 2:0:0:0: CD-ROm COMPAQ CD-ROM SC-140C B101 PQ: 0 ANSI: 5 [ 301.167466] Driver 'sr' needs updating - please use bus_type methods [ 301.171407] sr0: scsi3-mmc drive: 1x/48x cd/rw xa/form2 cdda tray [ 301.171507] Uniform CD-ROM driver Revision: 3.20 [ 301.181687] sr 2:0:0:0: Attached scsi generic sg0 type 5 Done. Check root= bootarg cat /proc/cmdline or missing modules, devices: cat /proc/modules ls /dev ALERT! /dev/cciss/c0d0p2 does not exist. Dropping to a shell! BusyBox v1.1.3 (Debian 1:1.1.3-5ubuntu7) Built-in shell (ash) Enter 'help' for a list of built-in commands. (initramfs) cat /proc/modules ... cciss 62378 2 - Loading 0xf8893000 ... pata_serverworks and ohci_hcd is also loading.
Normally when we a see at this point it indicates an interrupt routing issue. Did anything change in ACPI in the 2.6.24.2 kernel that was undone by the time .14 came out?
Since nothing has changed in the cciss module I guess it must be the ACPI causing trouble. Look at the IRQ and PCI Interrupt numbers below. Working kernel: ACPI: PCI Interrupt 0000:02:01.0[A] -> GSI 16 (level, low) -> IRQ 19 cciss0: <0x46> at PCI 0000:02:01.0 IRQ 19 using DAC Not working kernel: ACPI: PCI Interrupt 0000:02:01.0[A] -> CSI 16 (level, low) -> IRQ 16 cciss0: (0x46) at PCI 0000:02:01.0 IRQ 223 using DAC How did IRQ 16 suddenly become IRQ 223 in the 2.6.24 kernel?
Sorry for the late response. The kernel initially assigned us IRQ 16. When we enabled MSI-X on the controller we were assigned MSI-X vector 223. That's expected behaviour.
Strange, could this hang be caused by old firmware?
It's possible but not very likely. That said, we always recommend using the latest firmware available on hp.com.
It seems lika a firmware upgrade did the trick. 2.6.23 -> 25 now boots perfectly.